Вы находитесь на странице: 1из 28

Piketty and Marginal Productivity Theory:

A Superficial Application of a Very Bad Theory


by Fred Moseley
Mount Holyoke College
fmoseley@mtholyoke.edu
December 2014

The production function has been a powerful instrument for miseducation.


Joan Robinson (1953-54, p. 81)

When it comes to the neo-classical theory of income distribution, you are


either in or out with regard to the concept of marginal product
However, if marginal product is an incoherent or unusable concept, most
of neo-classical economics (including its approach to income distribution)
disintegrates; hence, the unwillingness to question marginal product
analysis.
Thomas Palley (2007; emphasis added)

Piketty and Marginal Productivity Theory:


A Superficial Application of a Very Bad Theory
This paper focuses on Thomas Pikettys explanation of the increased capital share
of total income in major economies since the early 1980s presented in his blockbuster
book. (Piketty 2014a, Chapter 6) Pikettys explanation is presented in terms of the
theoretical framework of the marginal productivity theory of distribution. The first
section of this paper critically reviews the essential elements of marginal productivity
theory; the second section summarizes Pikettys explanation of the increasing capital
share in terms of marginal productivity theory; the third section critically evaluates
Pikettys explanation; and the fourth section briefly presents an alternative heterodox
explanation of the increased capital share in recent decades. An Appendix briefly
criticizes Pikettys interpretation of Marxs theory of the falling rate of profit.
I should make it clear at the outset that I consider the empirical work done by
Piketty and his collaborators on income distribution to be a major contribution. This
empirical work has irrefutably documented the sharp increase in wealth and income
inequality in all major economies in recent decades. My criticisms of Pikettys book
have to do with his theoretical explanation of the increase in the capital share of income.
Another important contribution of Pikettys work is that it has helped to put the
question of the distribution of income back on the theoretical agenda of mainstream
economics, after a century of neglect. Of course, capitalism itself in recent decades, with
its sharply rising inequality, is mainly responsible for this increased interest in income
distribution; but Pikettys book has helped to galvanize this interest within mainstream

economics. This heightened interest in distribution presents an opportunity for heterodox


economists to engage in dialog and debate with the mainstream on this important issue.

1. Marginal productivity theory: It all depends on technology1

1.1 Production function


The foundation of marginal productivity theory is the production function. A
production function is a physical relation between quantities of inputs and a quantity of
output, without prices; it is a technical relation, like an industrial engineering plan.
A production function has the mathematical form Q = f(K, L, M), where K stands for
capital goods (buildings and equipment), L for labor, and M for material inputs (and
intermediate inputs in general).
It is obvious straightaway that an aggregate production function for an economy
as a whole, with single quantities of K, L, and M, does not exist in reality. Most
obviously, a single physical quantity of K does not exist, because it is not possible to add
together the many thousands (millions?) of different types of buildings and equipment,
each measured in terms of its own physical unit, in order to obtain a single physical
quantity of aggregate capital. The prices of the inputs cannot be used to measure the
quantity of the inputs because marginal productivity theory is supposed to explain the
prices of the inputs. A theory of inputs prices cannot take input prices as given; that
would be circular reasoning. Joan Robinson made this criticism a long time ago. This
aggregation problem in marginal productivity theory is widely recognized and
1

See Moseley 2012a, 2012b, and forthcoming for previous critiques of marginal
productivity theory.
3

acknowledged by many economists, and yet an aggregate production function continues


to be used (without a justifying discussion) in marginal productivity analyses of
economy-wide income distribution, including by Piketty.2

1.2 Marginal products


Leaving aside this insoluble aggregation problem, there are more insoluble
problems in marginal productivity, starting with the fundamental concept of marginal
product. A marginal product is the extra output that results from an increase of one unit
of one of the inputs holding all other inputs constant. Mathematically, the marginal
product is a partial derivative of the production function with respect to the changing
input; and a necessary condition for the existence of partial derivatives is that the
independent variables (e.g. K, L, and M) must themselves be mutually independent, so
that an increase in one of the variables does not cause or require an increase (or decrease)
in another variable.
However, in many production processes, this necessary condition is not satisfied;
i.e. it is not possible to increase output by increasing one of the inputs and holding all
other inputs constant. In these cases, marginal products and partial derivatives do not
exist even at the level of the individual firm, and marginal productivity theory cannot be
used to derive the demand for factors and to determine factor prices and income shares.

The theoretically invalid aggregate production function is also widely used other fields
of economics, most notably macroeconomic growth theory, development economics and
economic history.
4

Fixed proportions
The best known disqualifier for the existence of marginal products is
production processes with fixed proportions between the inputs, usually discussed as
fixed proportions between K and L. In many production processes, machines must be
used with a fixed number of workers; adding an extra machine does not increase output
unless an extra worker is also added to run the extra machine. Miller (2000) discusses
and cites a wide range of empirical studies (including by Edwin Mansfield, Joe Bain, A.A.
Walters, the NBER, and joint studies by the Fed and the Census Bureau) that generally
conclude that capital and labor in US manufacturing are usually employed in fixed
proportions:
In short, manufacturing firms generally adjust output in the short run by
increasing or decreasing the time in which capital and labor are used together.
Fixed proportions seem more suited to describing short-run manufacturing
processes than do variable proportions. (Miller 2000, p. 123)
However, fixed proportions means that the marginal products of capital and labor cannot
be separated, and thus the demand for capital and the demand for labor cannot be derived
independently and the prices of capital and labor cannot be determined independently.

Material inputs
A lesser known, but equally devastating disqualifier for the existence of
marginal products is the existence of material inputs in goods-producing industries (the
M in production functions). In these industries, it is not possible to increase output by
increasing capital (or labor) without also increasing material inputs (e.g. it is not possible
to produce another car without adding more windows, tires, etc.). In this case, K (or L)
and M are not mutually independent. An increase of K (or L) requires a complementary

increase of M in order to produce more output, and thus the necessary condition for the
existence of marginal products is not satisfied. It is another kind of fixed proportions,
except that in this case it is not a necessary fixed proportion between two inputs (capital
and labor), but rather a necessary fixed proportion between material inputs and output.
The main way that the problem of material inputs has been dealt with especially
in empirical work has been to assume away the problem, i.e. to assume that the
production functions are value added production functions, without material inputs.
However, this attempted solution does not work, because a production function is a
physical concept and value added is a nominal price concept the difference between the
price of the output and the prices of intermediate inputs. Prices and nominal value added
do not exist in engineering production plans. One can subtract the price of material
inputs from the price of the output to calculate nominal value added, because both prices
are in nominal terms which are commensurable. However, one cannot subtract the
physical quantity of material inputs from the physical quantity of output, because
materials and output in a given firm are different kinds of physical goods which are not
commensurable (e.g. what is the value added product of a car? a car without wheels?).
There is no common unit of measure in terms of which this subtraction of physical
quantities of inputs and outputs could be made. Therefore, a value added production
function is an oxymoron.
A more sophisticated attempt to solve the problem of material inputs is to assume
that the production function is separable, in such a way that capital and labor are
separable from material inputs, which allows for sequential optimization first materials
are held constant and firms optimize quantities of K and L to produce net output

(sometimes called real value added), and then firms optimize quantities of materials
and net output to produce gross output (Berndt and Christensen 1973, Arrow 1985,
Frondel and Schmidt 2004). However, the condition for separability is that the marginal
rate of substitution between capital and labor must be independent of the quantity of
materials (first articulated by Leontief 1947 and often called the Leontief condition),
which in turn requires that the production functions must be twice differentiable (because
the marginal rate of substitution is a ratio of partial derivatives of capital and labor; see
below). But production functions with material inputs are not even differentiable once.
Therefore the condition for separability is obviously not satisfied in production functions
with material inputs, and separability does not solve the problem of material inputs in
marginal productivity theory.3
Another sophisticated attempt to solve the problems of fixed proportions and
intermediate inputs has been the argument that, even though some factor proportions are
fixed within industries, the same set of factors are used in different industries, and in
different proportions; so that in this case changes in factor prices will lead to changes in
relative output prices and thus to changes in the demands for output, which feed back to
changes in the demand for inputs. In this way, substitution via consumption can play

Arrow 1985 has said that real value added is a latent concept which cannot be directly
observed. I would go further and say that real value added in physical terms is a does not
exist. Arrow continued: Without the separability assumption, however, it is hard to
assign any definite meaning to real value added, and probably the best thing to say is that
the concept should not be used when capital and labor are not separable from materials in
production. (p. 458; emphasis added) However, Arrow seems to have forgotten that
capital and labor are generally not separable from materials in production functions
because production functions with material inputs are not twice differentiable. Thus,
following Arrows advice, we should not use the concept of real value added or net
output. Value added cannot be reasonably deflated because it is not the product of a price
and an existing quantity of output.
7

the same role as input substitution within industries, and downward-sloping factor
demand curves can be derived. This argument was pioneered by Cassel in 1924.4
However, Stakelbergs critique of Cassel showed that the necessary condition for
this substitution via consumption to yield determinant input prices (sometimes called
the Cassel Condition) is that every set (or combination) of n inputs that are used in fixed
proportions within industries must also be used in at least n different industries.
Otherwise, some of the input prices will not be determined. However, once all the
specific machines and equipment and intermediate inputs used in fixed proportions in
many modern production processes are specified, it is highly unlikely that this stringent
condition will be fulfilled. Therefore, this more sophisticated defense of marginal
products and marginal productivity theory is also a failure.
Furthermore, even if the Cassel condition is miraculously satisfied for all sets of
fixed proportion inputs in the economy, it would still not be true that the prices of these
inputs are equal to (or determined by) their marginal products, because the marginal
products of these fixed proportion inputs do not exist. At best (i.e. assuming the Cassel
condition is satisfied), one could derive an inverse relation between the prices of inputs
and the demand for these inputs; but one cannot say that the prices of inputs are
determined by their marginal products. Marginal productivity theory continues to be
invalidated by the non-widespread existence of marginal products.

This summary of Cassels argument and Stakelbergs critique (next paragraph) is based
on Mandler 1999, Chapter 2.
8

1.3 Diminishing returns


Another important concept in marginal productivity theory, and in Pikettys
explanation of the increasing capital share, is the law of diminishing returns and
diminishing returns to capital in particular, which means that the marginal product of
capital will diminish over successive increases of one unit of capital, holding all other
inputs constant. This law of diminishing marginal products is necessary in marginal
productivity theory for the derivation of the demand for capital (and labor) as downwardsloping curves. However, we have seen above that marginal products do not exist in
many production processes; therefore diminishing marginal products also do not exist
and this so-called law is in fact not an empirical law.
Diminishing returns was first introduced by the classical economists who usually
had agriculture in mind. Diminishing returns made some sense in 18th and early 19th
century agriculture adding workers to a fixed quantity of land with a fixed amount of
seed although even in this case an additional worker would probably require an
additional hoe or some other agricultural tool in order to produce additional output. But
diminishing returns does not make any sense in modern industry in which machines and
workers are usually combined in fixed proportions and also additional output usually
requires additional material inputs.
In his review of Pikettys book, Lawrence Summers stated:
Economists universally believe in the law of diminishing returns. As capital
accumulates, the incremental return on an additional unit of capital declines.
(Summers 2014; emphasis added)
However, Summers and mainstream economists in general either do not realize that
marginal products often dont exist in many production processes (i.e. the incremental

return of an additional unit of capital is zero unless accompanied by additional labor


and/or additional material inputs) or (more likely) they choose to ignore this
inconvenient truth and maintain their belief in diminish returns.
It should also be emphasized that diminishing returns assumes that technology
remains constant (i.e. a production function characterized by diminishing returns is given
and constant). Diminishing returns is a short-run concept that does not apply over
decades in which technology is changing significantly. According to marginal
productivity theory, technological change is supposed to offset diminishing returns, so
that returns generally do not diminish. With technological change, the return to capital
should increase, not decrease. The assumption of constant technology makes the concept
of diminishing returns even more unrealistic and inapplicable to real capitalist economies
in which technological change is pervasive and continuous.

1.4 Elasticity of substitution


Another important concept in marginal productivity theory, and in Pikettys
explanation of the increasing capital share, is the elasticity of substitution between
capital and labor, which is defined in terms of the marginal products of capital and labor
and thus the non-existence of marginal products also applies to the elasticity of
substitution. The elasticity of substitution is a measure of the curvature of an isoquant
(which shows all the different combinations of K and L that can be used efficiently to
produce a given level of output).
The slope of an isoquant is the rate of technical substitution (RTS), and the RTS
of capital for labor is equal to the ratio MPL/MPK. Moving along an isoquant both K/L

10

and the RTS change (the latter due to diminishing returns). The elasticity of
substitution () is the relation between the percentage change in K/L that occurs in
response to a given the percentage change in the RTS, which can be expressed in terms of
natural logs as follows:
(1)

K for L = d ln(K/L) / d ln RTS = d ln(K/L) / d ln (MPL/MPK)

However, once again, since marginal products do not exist in many production processes,
neither does the rate of technical substitution and thus neither does the elasticity of
substitution between K and L
It is argued further that in perfect competition equilibrium, the ratio of marginal
products is equal to the ratio of factor prices:
MPL/MPK = PL/ PK
Thus the elasticity of substitution at the equilibrium point on an isoquant can be
reformulated as:
(1)

K for L = d ln(K/L) / d ln (PL/PK)

This formulation of the elasticity of substitution is used to analyze and estimate the
substitution of inputs in response to a change in their relative prices. However, this
formulation still assumes in the background that marginal products exist (but they often
do not) and also adds the unrealistic assumption of perfect competition.
In summary, marginal productivity theory is a very unrealistic theory which
cannot reasonably be used to analyze the distribution of income in the real world.
Marginal products often do not exist, which means that diminishing returns and the
elasticity of substitution also do not exist, and the whole theory falls apart and lacks a
coherent logical foundation. In addition, the concept of diminishing returns assumes that

11

technology remains constant and thus cannot be used of explain trends in the distribution
of income in real economies in which technology is constantly changing.
Tom Palley has commented in a recent blog that mainstream economists have
decided to ignore the logical incoherence of marginal productivity theory. (Palley
2014) But this is not acceptable scholarly and scientific practice. We should not let
mainstream economists get away with this scientific malpractive.

2. Pikettys explanation of the increase in the capital share: Everything depends


on technology
Pikettys explanation of the significant increase in the capital share in major
economies in recent decades is based on marginal productivity theory, but it is not a
rigorous application of the theory, as we shall see. The key equation in Pikettys
explanation of the increase in the capital share is his First Fundamental Law of
Capitalism, which is a decomposition of the capital share into its two accounting factors:
(2)

/Y = (/K) (K/Y)5

where /Y is the capital share (i.e. the share of profit in total income), /K is the rate of
return to capital (i.e. the rate of profit, the ratio of total profit in the economy to the total
capital invested), and K/Y is the capital-income ratio (or the capital-output ratio).
According to Pikettys estimates, the K/Y ratio generally increased in the major
economies, and his explanation for this increase is that the rate of return on capital was
greater than the rate of growth of output (the now famous inequality r > g); this
conclusion is based on a modified Howard-Domar-Solow growth model that will not be
5

Piketty expresses this equation in terms of Greek letters which stand for these ratios:
= r . I find it clearer to express the equation in terms of the ratios themselves.
12

examined here. According to equation (2), this increase in the K/Y ratio by itself had a
positive effect on the capital share of income.
Piketty then invokes marginal productivity theory and assumes that the rate of
return to capital is determined by the marginal productivity of capital and further assumes
diminishing returns to capital to argue that this increase in the K/Y ratio caused the
marginal productivity of capital to decline, which in turn caused the rate of return on
capital to decline. Piketty does not provide an explanation or justification of the
assumption of diminishing returns, but just asserts it (too much capital kills the rate of
return; 215-16). He simply states that it is natural to expect that the marginal
productivity of capital decreases as the stock of capital increases. (215) He gives an
example of agriculture and oddly holds the number of workers constant while increasing
the quantity of land (the usual assumption is the other way around), and argues that it is
likely that the extra yield [the marginal product of land] of an additional hectare of land
will be limited (Pikettys definition of capital includes land). Another example is
residential housing (Pikettys definition of capital also includes residential housing; see
below for further discussion) whose product is well-being, and he argues that if a
country has already built a huge number of new dwellings, then the increase to wellbeing of one additional building would no doubt be very small. (215) The only
comment about modern capitalist industry is the next sentence which asserts: The same
is true for machinery and equipment of any kind: marginal productivity decreases with
quantity beyond a certain threshold. (215)
Piketty does not mention and does not seem to be aware that diminishing returns
in marginal production theory assumes that technology remains constant and thus cannot

13

explain the decline in the rate of return to capital in recent decades in which there has
been significant technological change.
In any case, according to Pikettys equation (2), the reduction in the rate of return
to capital by itself had a negative effect on the capital share. Therefore, the net effect of
these contradictory changes on the capital share depends on the relative rates of change of
the increase in the K/Y ratio and the decrease in the rate of return to capital.
Again invoking marginal productivity theory, Piketty argues further that these
relative rates of change depend in turn on the elasticity of substitution of capital for
labor (216-17), which he defines in terms of equation (2) as the ratio of the percentage
change in K/Y to a given percentage change in the rate of return to capital:
(3)

*K for L = d ln (K/Y) / d ln (r)

(We will discuss below the differences between Pikettys definition of the elasticity of
substitution and the standard definition in marginal productivity theory in equation (1)
above). According to this definition and equation (2), if the elasticity of substitution
is > 1, then the capital share will increase as a result of these changes, because firms
respond to the lower rate of return on capital by replacing labor with capital on a more
significant scale.
Piketty argues that the elasticity of substitution in less developed economies is in
general < 1, and it increases along with development, so that it is in general > 1 in more
advanced economies. Technological advancement means that there are more plentiful
and more profitable possibilities for substituting machines for labor, and that firms will
do so on an increasing scale. Piketty also argues that historical estimates of the variables

14

in equation (3) for advanced economies suggest that the elasticity of substitution for these
economies is in a range 1.3 and 1.6. (220-21)
In sum, according to Piketty (using marginal productivity theory), the increase in
the capital share in recent decades was caused by a combination of an increase in the K/Y
ratio, diminishing returns, and an elasticity of substitution greater than one, which meant
that the increase in the K/Y ratio was greater than the decline in the rate of return to
capital, thus resulting (according to equation 2) in an increase in the capital share of
income. As Piketty put it: Everything depends on the vagaries of technology. (216;
see also The Caprices of Technology (234); emphasis added) According to this
marginal productivity explanation, if the elasticity of substitution in recent decades had
been less than one, rather than greater than one, then the capital share of income would
have decreased, not increased.

3. Critique of Pikettys explanation: a superficial application of a bad theory


Pikettys explanation of the increase in the capital share accepts marginal
productivity theory at face value and without mentioning any of the well-known
problems discussed in Section 1: the aggregation problem (Piketty assumes an aggregate
production function without comment); fixed proportions between capital and labor and
fixed proportions between material inputs and output in many production processes, both
of which render the fundamental concept of the marginal product of capital (or labor)
invalid, which in turn implies that the elasticity of substitution between capital and labor
does not exist in the real world; and the unrealistic assumption of constant technology in
the concept of diminishing returns. Because of all these unaddressed problems, Pikettys

15

explanation of the increasing capital share in terms of marginal productivity theory is a


non-starter such a problematic theory cannot be accepted as the basis for a valid
explanation.
Not only does Piketty assume an aggregate production function, but he also
assumes a value added aggregate production function, which (as discussed above) is a
contradiction in terms. Although an aggregate value added production function is widely
used by neoclassical economists, it is logically incoherent and the form of marginal
productivity theory that has the least theoretical foundation (that is to say, none).
Although Piketty employs the concepts and logic of marginal productivity theory
(production function, marginal product of capital, diminishing returns, elasticity of
substitution) in his explanation of the increase in the capital share, the variables in
Pikettys equation (1) and his explanation of the increase in the capital share are in fact
not the same variables in marginal productivity theory. Piketty does not discuss these
differences and does not seem to be aware of them.
To begin with (as just discussed), all the variables in equation (1) are nominal
variables, in terms of prices, rather than physical variables as in the production function
of marginal productivity theory. Pikettys K is in terms of the price of capital goods,
rather than the quantity of capital goods. Thus, an increase in the price of capital goods
would by itself (without any change in the physical ratio) increase Pikettys nominal K
and nominal K/Y ratio, but would not increase the standard physical K/L ratio. Similarly,
Pikettys Y is in terms of the price of output [nominal income], rather than the quantity of
output. (From now on, I will use K* and Y* to stand for Pikettys nominal capital and
nominal income). Piketty argues that nominal variables are more relevant for income

16

distribution than physical variables. But according to marginal productivity theory,


nominal income variables are supposed to be determined by the physical properties of the
physical production function.
Ronglie (2014) has discussed this difference between Pikettys nominal K* and
the standard neoclassical physical K and has argued that increases in the relative price of
capital (i.e. capital gains) accounted for 84% of the increase in Pikettys nominal
K*/Y* in the eight countries in Pikettys sample. (p. 15) Thus very little of Pikettys
significant increase in the K*/Y* ratio was due to the quantity of physical capital goods,
which is what marginal productivity theory is supposed to be about.
Another problem with Pikettys definition of capital is that he equates capital
with wealth which also includes residential housing. Since most of housing is owneroccupied, it does not belong in a production function for capitalist firms. According to
Rognlie, if the value of housing is excluded from the definition and estimates of capital,
this correction alone eliminates 80% of the increase in the capital-income ratio in the
eight countries in Pikettys sample during this period. Thus very little of Pikettys
significant increase in the K*/Y* ratio was due to an increase in the buildings and
equipment utilized by capitalist firms to produce output, which again is what marginal
productivity theory is supposed to be about.6

Solow (2014) has noted this difference between Pikettys broad definition of
capital/wealth and the neoclassical concept of capital as a factor of production, but he
calls it a small ambiguity which should not affect the long-run trends in the
capital/income ratio. But Solow should look again at the estimates.

17

In addition, the price variable in Pikettys equation (1) is different from the usual
price variables in marginal productivity theory. The price of capital goods in marginal
productivity theory (PK) is a unit price, the price per unit of capital goods (whatever that
unit might be); but the rate of return to capital (r) in Pikettys equation (1) is not the unit
price of capital goods, but is instead the ratio of two aggregate nominal prices: the total
profit in the economy as a whole divided by the total capital invested.
There is also a logical problem with Pikettys definition of the rate of return to
capital. The elasticity of substitution is supposed to measure the response of capitalist
firms to a change in the relative unit prices. PK is a cost to firms, the cost that firms have
to pay to purchase (or rent) capital goods; a reduction in PK is supposed to induce firms to
substitute cheaper capital goods for labor. However, the aggregate ratio in Pikettys
equation is a profit variable, not a cost variable. It does not make sense that a reduction
in this profit ratio would induce firms to substitute capital goods for labor. In this case,
capital goods have not become cheaper, but rather have become less profitable.
We can see from these differences that Pikettys definition of the elasticity of
substitution (equation 3) is also different from the standard neoclassical definition
(equation 1). Both the numerators and the denominators in these respective definitions
are different. The numerator in the standard definition is the K/L ratio in physical terms,
whereas the numerator in Pikettys definition is the K*/Y* ratio in nominal terms. Thus
(as mentioned above), an increase in the price of capital goods by itself (without any
change in the physical ratio) would increase Pikettys nominal K* and nominal K*/Y*
ratio, but would not affect the standard physical K/L ratio. In the denominator, the ratio
of the standard elasticity of substitution is the ratio of marginal products (MPL/MPK)

18

which in perfect competition equilibrium is supposed to be equal to the ratio of relative


unit prices (PL/PK). On the other hand, the denominator of Pikettys definition of the
elasticity of substitution is a ratio of aggregate magnitudes (total profit / total capital).
A number of commentators have criticized Pikettys estimates of the elasticity of
substitution (between 1.3 and 1.6, as discussed above), and have argued that the
consensus of the empirical literature on estimates of the elasticity of substitution in the
US economy is that it is much lower than 1 (the critical value). In a review article,
Chirinko (2008) concludes that most estimates are in the range of 0.4 to 0.6. Most of the
critics have argued that the main problem with Pikettys estimates is that the rate of
return to capital (in the denominator) is defined as a net concept (net of the depreciation
cost of capital goods) and the appropriate definition should be a gross concept (including
depreciation cost). However, no one (that I have seen) has discussed the more
fundamental difference between Pikettys rate of return to capital as an aggregate price
ratio and the standard price of capital goods as a unit price. The standard empirical
literature tries to estimate the responsivness of the physical K/L ratio to changes in the
relative unit prices of K and L (PK/ PL). Pikettys estimates, on the other hand, are
calculated from the nominal estimates of K*/Y* and the rate of return to capital (/K) in
equation (2), and are supposed to estimate the responsiveness of K*/Y* to the rate of
return. It is not surprising that the estimates of these two very different definitions of the
elasticity of substitution have such different magnitudes.
In sum, Pikettys explanation of the increase in the capital share in recent decades
is a flawed application of a very bad theory. Piketty seems to accept marginal
productivity theory, but marginal productivity theory is not a logically coherent theory,

19

especially with a value added aggregate production function. Marginal products do not
exist in many production processes and diminishing returns is based on the unrealistic
assumption of constant technology.
Furthermore, even though Piketty seems to accept marginal productivity theory
and employs the concepts of marginal productivity theory, it is a very superficial and
non-rigorous application of marginal productivity theory, as discussed in this section.
These fundamental differences essentially leaves Pikettys explanation of the increased
capital share in recent decades without any theoretical foundation at all and reduces his
explanation to a set of assertions about the aggregate nominal ratios in equation (2),
based mainly on extrapolation from recent past trends.
One could say perhaps that Piketty is just using marginal productivity theory (or a
bastardized version of it) to demonstrate to mainstream economists that, even within their
own theory, there is no natural tendency for income shares to remain constant over time
(constant shares have been believed by many economists for a long time). I think there
may be something to that speculation. But if true, it is a bad tactical decision. One does
not need marginal productivity theory (or any other theory) to demonstrate that incomes
shares do not remain constant over time; all one needs to do is look at the data. And then
the task is to provide the best possible explanation of the increase in the capital share in
recent decades, which would provide insights into the types of government policies that
would be the most effective in offsetting the increase in the capital share. One could still
demonstrate that within marginal productivity theory there is no natural tendency for
income shares to remain constant without endorsing the theory; and one could also
discuss at least some of the well-known criticisms of marginal productivity theory, and at

20

the same time present an alternative theory which would provide a better explanation of
the increase in the capital share in recent decades. By uncritically employing marginal
productivity theory to (try to) explain the increase in the capital share, Piketty reinforces
the hegemony of marginal productivity theory, which is a major piece of capitalist
ideology; and, worst of all, it does not provide a valid explanation of this important
phenomenon.
Therefore, if we want to understand the underlying causes of the increasing
capital share in recent decades we have to look elsewhere besides Piketty and marginal
productivity theory. And in fact we dont have to look far. The next section will briefly
discuss a heterodox explanation of the increasing profit share presented in various forms
by a number of authors, and based on the increasing economic and political power of
capitalists over workers in recent decades (e.g. Dumnil and Levy 2011, Mishel et al
2013, Bernstein and Baker 2013, Kotz 2014).

4. Heterodox theory of the increased profit share: economic and political power
There is an alternative and much more persuasive heterodox explanation of the
increase in the profit share (capital share) in advanced economies in recent decades. The
profit share is equal to 1 minus the wage share, and this heterodox theory usually focuses
on the wage share. According to this heterodox theory, the wage share depends mainly
on the balance of power between capitalists and workers economic power and political
power. If the balance of power shifts away from workers toward and capitalists, then the
wage share will decline and the profit share will increase, and vice versa. The balance of
economic power between capitalists and workers in turn depends primarily on: (1) the

21

rate of unemployment and the threat of unemployment, (2) the mobility of capital, and (3)
the prevalence of unions. The higher the rate of unemployment or the threat of
unemployment and the greater the mobility of capital, the greater will be the power of
capitalists over workers and the lower will be the wage share of income. The existence
of unions increases the countervailing power of workers. Political power can be used to
influence labor laws and especially the minimum wage. Economic power is translated
into political power, which in turn protects and enhances economic power, all of which
leads to an increasing profit share.
In the US economy in recent decades, all of these factors have contributed to the
observed decline in the wage share and increase in the profit share. The rate of
unemployment since 1970 has been generally higher than in the early postwar period, and
the threat of unemployment has been much greater due to globalization and the enhanced
mobility of capital, and these factors have weakened the power of workers and increased
the power of capitalists in the conflict over wages. The percentage of the labor force that
are union members has declined sharply from 29% in 1975 to 11% today. In addition,
capitalists have used their increased political power (since Reagan) to weaken labor laws
(e.g. right to work laws) and to block minimum wage increases, which has resulted in a
25% decline in the real minimum wage since 1970, which contributed significantly to the
declining wage share.
I argue that this heterodox explanation of the declining wage share and increasing
profit share based on economic and political power is much more realistic and persuasive
than Pikettys explanation based on logically incoherent marginal productivity theory and
non-existent marginal products of a non-existent aggregate value added production

22

function and the implicit assumption of constant technology. Contrary to Pikettty,


income shares do not depend primarily on technology, but instead depend primarily on
economic and political power.
The policy recommendations that follow from this heterodox explanation of the
declining wage share include: expansionary fiscal and monetary policy to reduce
unemployment, a significant increase in the minimum wage so that the real wage is
increased at least to the level of the 1970s, and more favorable labor laws to enable union
organization. Pikettys policy recommendation of a tax on wealth (especially inherited
wealth) is also a good idea, but it does not address the root causes of the decline of the
wage share and the increase of the profit share.

23

Conclusion
I conclude that the profit and wage shares of income are not determined by
technology (marginal products, diminishing returns, elasticity of substitution), but are
instead determined by the balance of power and the class conflict between capitalists and
workers. If the wage share is to be increased in the years ahead, then the working class
and its allies will have to organize better and exert more economic and political power in
this ongoing class conflict with capitalists over wages and the distribution of income.

24

Appendix: Pikettys ignorance of Marxs theory of the falling rate of profit


Piketty makes the outrageous claim that Marxs theory of the falling rate of profit
neglected technological change and is based on the assumption that technology remains
constant!
Like his predecessors, Marx totally neglected the possibility of durable
technological progress and steadily increasing productivity (10)
That is, Piketty interprets Marxs theory of the falling rate of profit as similar to the
diminishing returns in marginal productivity theory. But Pikettys interpretation is
completely wrong and is evidence of Pikettys ignorance of Marxs theory. Marxs
theory of the falling rate of profit is focused precisely on technological change and the
effects of technological change on the rate of profit.
Marx argued that technological change tends to be labor-saving, and since (according
to the labor theory of value) labor is the source of profit, labor-saving technological
change causes the rate of profit to fall for the economy as a whole (even though profit
increases for the innovative capitals, at least temporarily). From Chapter 13 of Volume 3
(The Law as Such):
The progressive tendency for the general rate of profit to fall is thus
simply the expression, peculiar to the capitalist mode of production, of the
progressive development of the social productivity of labour.
(C.III. 319; italics emphasis in the original, bold emphasis added)
I have discussed this key aspect of Marxs theory of the falling rate of profit (caused by
labor-saving technological change) in detail in a recent paper (Moseley 2014).
Piketty admitted in an interview in the New Republic (entitled I Dont Care for
Marx) that he has never really studied Marxs theory because it is so difficult, and he

25

even sought sympathy from the interviewer because of its difficulty. An excerpt from the
interview went like this:
Interviewer: Can you talk a little bit about the effect of Marx on your thinking
and how you came to start reading him?
Piketty: Marx?
Interviewer: Yeah.
Piketty: I never managed really to read it. I mean I dont know if youve tried
to read it. Have you tried? (Piketty 2014b; emphasis added)
But if this is true, then Piketty should not have said anything about Marxs theory of the
falling rate of profit in his book, and he would not have made such an egregious error.
The irony is that Pikettys own explanation of the decline in the rate of return to
capital is based on marginal productivity theory and its law of diminishing returns, and
this law does assume constant technology. Piketty (and marginal productivity theory in
general) is guilty of what he accuses Marx of! On the other hand, Marxs theory of the
falling rate of profit is not based on marginal products and diminishing returns, and does
not assume constant technology; but is instead based on the labor theory of value and
analyzes at great length the effects of technological change on the rate of profit.

26

References
Arrow, Kenneth 1985. The Measurement of Real Value Added, in Collected Papers of
Kenneth J. Arrow, Volume 5, Production and Capital, Cambridge MA: Belknap
Press for Harvard University Press.
Berndt and Christensen 1973. The Internal Structure of Functional Relationships:
Separability, Substitution, and Aggregation, Review of Economics and Statistics,
40: 403-10.
Bernstein, Jared and Dean Baker 2013. Getting Back to Full Employment: A Better
Bargain for Working People, Washington: Center of Economic and Policy Research.
Cassel, Gustav 1924. The Theory of Social Economy, translated by Joseph McGabe,
New York: Harcourt, Brace and Company.
Dumnil, Gerard and Dominique Levy 2011. The Crisis of Neoliberalism, Cambridge
MA: Harvard University Press.
Frondel and Schmidt 2004. Facing the Truth about Separability: Nothing Works
Without Energy, Ecological Economics, 51: 217-23.
Kotz, David 2014. The Rise and Fall of Neoliberal Capitalism, Cambridge MA:
Harvard University Press.
Mandler, Micnael 1999. Dilemmas in Economic Theory. New York: Oxford University
Press.
Miller, Richard A. 2000. Ten Cheaper Spades: Production Theory and Cost Curves in
the Short Run, Journal of Economic Education, 31, 2: 119-30.
Mishel, Lawrence, Josn Bevins, Elise Gould, and Heidi Shierholz 2013. State of Working
America, 12th edition, Ithaca: ILR Press.
Moseley, Fred 2012a. A Critique of the Marginal Productivity Theory of the Price
of Capital, Real World Economics Review, 59: 131-137.
Moseley, Fred 2012b. Mankiws Attempted Resurrection of Marginal Productivity
Theory, Real World Economics Review, 61: 115-124.
Moseley, Fred 2014. The Development of Marxs Theory of the Falling Rate of Profit in the
Four Drafts of Capital, paper presented at the conference Marxs Capital and the
MEGA, International Institute of Social History, Amsterdam.

27

Moseley, Fred forthcoming. The Marginal Productivity Theory of Capital in


Intermediate Microeconomics Textbooks: A Critique, Review of Radical Political
Economics.
Palley, Thomas 2007. Marginal Productivity Theory and the Mainstream,
http://www.thomaspalley.com/?p=81.
Palley, Thomas 2014. The accidental controversialist: deeper reflections on Thomas
Pikettys Capital, Real World Economics Review, no. 67, pp. 143-46.
Piketty, Thomas 2014a. Capital in the Twenth-First Century, translated by Arghur
Goldhammer. Cambridge MA: The Belknap Press of Harvard University Press.
Piketty, Thomas 2014b. Thomas Piketty: I Dont Care for Marx, New Republic,
http://www.newrepublic.com/article/117655/thomas-piketty-interview-economistdiscusses-his-distaste-marx
Robinson, J. 1953-54. The Production Function and the Theory of Capital, Review of
Economic Studies, 21: 81-106.
Rognlie, Matthew 2014. A Note on Piketty and Diminishing Returns to Capital,
http://www.mit.edu/~mrognlie/piketty_diminishing_returns.pdf.
Solow, Robert 2014. Thomas Piketty is Right, New Republic, April 22.
Summers, Lawrence 2014. The Inequality Puzzle, Democracy, #33.

28